Speech Synthesis and Spanish Intonation: The Effects of Peak Alignment on the Perception of Sentence Type

By:
To add a paper, Login.

This paper presents a specific example of how intonation in language can be studied computationally using speech synthesis techniques. The study measures the effect of fundamental frequency and peak alignment on the perception of sentence type in Mexico City Spanish. Two one-word sentences were recorded and their intonational peaks and frequencies were systematically manipulated using a computer. Peaks were displaced in different intervals to the left and to the right, and the frequencies were adjusted to 270Hz and 320Hz. Then, two native speakers from Mexico City listened to these synthesized utterances and were asked to identify them as either declarative, interrogative or exclamative sentences. Results of the study show that interrogative sentences with the peak displaced to the right by 140ms are perceived as declaratives, and that declaratives with the peak displaced to the right by 100ms and 120ms are perceived as exclamative. Higher frequencies (320Hz) are perceived more often as interrogatives than as declaratives.


Keywords: Language, Technology, Intonation, Speech Synthesis, Spanish, Peak Alignment, Perception, Computational
Stream: Technology and Applied Sciences
Presentation Type: Virtual Presentation in English
Paper: Speech Synthesis and Spanish Intonation: The Effects of Peak Alignment on the Perception of Sentence Type


Mike Field

Graduate Student, Department of Spanish and Portuguese, University of Toronto
Toronto, Ontario, Canada

Mike Field is a graduate student of Hispanic Linguistics in the Department of Spanish and Portuguese at the University of Toronto. Currently in his first year of a doctoral program, his primary focus of study is Computational Linguistics and the examination of language from a technological perspective. His most recent projects include a computational study of bare nouns in Madrid Spanish, for use in grammar checking software and natural language generation systems. He has also carried out experiments using speech synthesis techniques that examine the intonation of Mexico City Spanish, for use in text-to-speech systems. He has an MA in Spanish from the University of Toronto and a B.A.H. in Spanish from Queen’s University at Kingston, Ontario.

Ref: I08P0964